The Culinary Creator: Your AI Sous Chef & Recipe Innovator

This is a submission for the Runner H "AI Agent Prompting" Challenge What I Built The Culinary Creator: Your AI Sous Chef & Recipe Innovator I built an autonomous AI agent, powered by Runner H, designed to be a personal culinary assistant, recipe innovator, and meal planner. This agent leverages Google Sheets to manage ingredients, dietary preferences, and meal inventory, and Google Docs to generate customized recipes, shopping lists, and cooking guides. The problem it solves is the daily dilemma of meal planning, food waste due to forgotten ingredients, and the creative block often encountered when trying to come up with new and exciting dishes. The Culinary Creator ensures delicious, personalized meals while optimizing ingredient usage and minimizing food waste. Demo How I Used Runner H My Runner H agent orchestrates a delightful and efficient culinary workflow, seamlessly integrating Google Sheets for data management and Google Docs for personalized recipe generation and meal planning, all guided by intelligent prompting techniques. Here\'s a breakdown of the workflow and how Runner H is utilized: 1. Pantry & Preference Management (Google Sheets & Few-shot Prompting) Runner H begins by establishing a comprehensive inventory of available ingredients and understanding the user\'s culinary preferences within interconnected Google Sheets: Pantry_Inventory.gsheet: Tracks ingredients (Item, Quantity, Unit, Expiration Date, Location). Dietary_Preferences.gsheet: Records dietary restrictions (e.g., Vegetarian, Gluten-Free), allergies, disliked ingredients, and preferred cuisines. Meal_History.gsheet: Logs past meals (Date, Recipe Name, Rating, Notes). Runner H assists in populating these sheets. For instance, when the user adds a new ingredient, Runner H, using Few-shot Prompting (drawing from examples of common pantry staples or ingredient categories), can suggest typical units or common expiration durations, which are then added to the Pantry_Inventory.gsheet. This ensures a structured and consistent database of available food items and preferences. 2. Dynamic Recipe Generation (Google Docs & Meta Prompting) Based on the available ingredients in Pantry_Inventory.gsheet and the preferences in Dietary_Preferences.gsheet, Runner H dynamically generates customized recipes in Google Docs: Personalized Recipes: Creates unique recipes tailored to the user\'s current inventory and taste. Ingredient Substitution Suggestions: Offers alternatives if a specific ingredient is missing. Cooking Guides: Generates step-by-step instructions, including preparation tips and cooking times. In this phase, Meta Prompting is crucial. Runner H is prompted to generate the optimal recipe format, complexity, and presentation style. For example, the user might prompt: "Generate a quick, healthy dinner recipe using chicken, broccoli, and rice, suitable for a weeknight, and make it Asian-inspired." Runner H then intelligently selects and synthesizes information, ensuring the recipe is delicious, practical, and aligns with the user\'s request, and writes it directly into a new Google Doc (e.g., "Asian_Chicken_Broccoli_Stir_Fry.docx"). 3. Smart Meal Planning & Shopping List Generation (Google Sheets & Active-Prompt) Runner H assists with intelligent meal planning. The user can request meal ideas for a week, and Runner H will suggest recipes based on pantry inventory and preferences, minimizing waste. If the user approves a meal plan, Runner H automatically generates a shopping list by comparing required ingredients with the Pantry_Inventory.gsheet. If the user requests a recipe that requires a missing ingredient, Runner H uses Active-Prompt to ask for clarification or suggest alternatives: "You\'ve requested a recipe for [Dish Name], but you\'re missing [Ingredient]. Would you like me to suggest a substitute, or add [Ingredient] to your shopping list?" This proactive interaction ensures efficient meal preparation and grocery shopping. 4. Culinary Experimentation & Adaptation (Google Docs & Chain-of-Thought Prompting) Runner H encourages culinary experimentation. If a user wants to adapt an existing recipe or explore a new flavor profile, Runner H can assist. It employs Chain-of-Thought (CoT) Prompting to reason through culinary principles. For example, if the user asks: "How can I make this pasta dish spicier without adding chili flakes?" Runner H might reason: "To increase spiciness, consider ingredients with capsaicin or allicin. Garlic, ginger, or certain herbs could work. What flavor profile are we aiming for?" and then suggest specific ingredients and quantities to add to the recipe in Google Docs. This iterative reasoning allows for creative recipe modifications. 5. Continuous Improvement & Palate Refinement (Reflexion) To continuously improve its recipe generation and meal planning capabilities, Runner H incorporates Ref

Jun 7, 2025 - 03:10

This is a submission for the Runner H "AI Agent Prompting" Challenge

What I Built

The Culinary Creator: Your AI Sous Chef & Recipe Innovator

I built an autonomous AI agent, powered by Runner H, designed to be a personal culinary assistant, recipe innovator, and meal planner. This agent leverages Google Sheets to manage ingredients, dietary preferences, and meal inventory, and Google Docs to generate customized recipes, shopping lists, and cooking guides. The problem it solves is the daily dilemma of meal planning, food waste due to forgotten ingredients, and the creative block often encountered when trying to come up with new and exciting dishes. The Culinary Creator ensures delicious, personalized meals while optimizing ingredient usage and minimizing food waste.

Demo

How I Used Runner H

My Runner H agent orchestrates a delightful and efficient culinary workflow, seamlessly integrating Google Sheets for data management and Google Docs for personalized recipe generation and meal planning, all guided by intelligent prompting techniques.

Here\'s a breakdown of the workflow and how Runner H is utilized:

1. Pantry & Preference Management (Google Sheets & Few-shot Prompting)

Runner H begins by establishing a comprehensive inventory of available ingredients and understanding the user\'s culinary preferences within interconnected Google Sheets:

Pantry_Inventory.gsheet: Tracks ingredients (Item, Quantity, Unit, Expiration Date, Location).
Dietary_Preferences.gsheet: Records dietary restrictions (e.g., Vegetarian, Gluten-Free), allergies, disliked ingredients, and preferred cuisines.
Meal_History.gsheet: Logs past meals (Date, Recipe Name, Rating, Notes).

Runner H assists in populating these sheets. For instance, when the user adds a new ingredient, Runner H, using Few-shot Prompting (drawing from examples of common pantry staples or ingredient categories), can suggest typical units or common expiration durations, which are then added to the Pantry_Inventory.gsheet. This ensures a structured and consistent database of available food items and preferences.

2. Dynamic Recipe Generation (Google Docs & Meta Prompting)

Based on the available ingredients in Pantry_Inventory.gsheet and the preferences in Dietary_Preferences.gsheet, Runner H dynamically generates customized recipes in Google Docs:

Personalized Recipes: Creates unique recipes tailored to the user\'s current inventory and taste.
Ingredient Substitution Suggestions: Offers alternatives if a specific ingredient is missing.
Cooking Guides: Generates step-by-step instructions, including preparation tips and cooking times.

In this phase, Meta Prompting is crucial. Runner H is prompted to generate the optimal recipe format, complexity, and presentation style. For example, the user might prompt: "Generate a quick, healthy dinner recipe using chicken, broccoli, and rice, suitable for a weeknight, and make it Asian-inspired." Runner H then intelligently selects and synthesizes information, ensuring the recipe is delicious, practical, and aligns with the user\'s request, and writes it directly into a new Google Doc (e.g., "Asian_Chicken_Broccoli_Stir_Fry.docx").

3. Smart Meal Planning & Shopping List Generation (Google Sheets & Active-Prompt)

Runner H assists with intelligent meal planning. The user can request meal ideas for a week, and Runner H will suggest recipes based on pantry inventory and preferences, minimizing waste. If the user approves a meal plan, Runner H automatically generates a shopping list by comparing required ingredients with the Pantry_Inventory.gsheet. If the user requests a recipe that requires a missing ingredient, Runner H uses Active-Prompt to ask for clarification or suggest alternatives: "You\'ve requested a recipe for [Dish Name], but you\'re missing [Ingredient]. Would you like me to suggest a substitute, or add [Ingredient] to your shopping list?" This proactive interaction ensures efficient meal preparation and grocery shopping.

4. Culinary Experimentation & Adaptation (Google Docs & Chain-of-Thought Prompting)

Runner H encourages culinary experimentation. If a user wants to adapt an existing recipe or explore a new flavor profile, Runner H can assist. It employs Chain-of-Thought (CoT) Prompting to reason through culinary principles. For example, if the user asks: "How can I make this pasta dish spicier without adding chili flakes?" Runner H might reason: "To increase spiciness, consider ingredients with capsaicin or allicin. Garlic, ginger, or certain herbs could work. What flavor profile are we aiming for?" and then suggest specific ingredients and quantities to add to the recipe in Google Docs. This iterative reasoning allows for creative recipe modifications.

5. Continuous Improvement & Palate Refinement (Reflexion)

To continuously improve its recipe generation and meal planning capabilities, Runner H incorporates Reflexion. After each meal, the user can rate the recipe in Meal_History.gsheet. Runner H implicitly evaluates the effectiveness of its generated recipes and suggestions. If a recipe receives a low rating or frequent modifications, Runner H self-corrects its internal logic and adjusts its prompting strategies for future recipe generation. For example, if a user consistently dislikes cilantro, Runner H will learn to avoid suggesting recipes with cilantro or offer alternatives. This iterative self-improvement ensures that the agent becomes an increasingly refined and personalized culinary assistant.

Use Case & Impact

Who Benefits:

This Runner H workflow is invaluable for:

Home Cooks: Simplifies meal planning, reduces food waste, and inspires new culinary creations.
Busy Professionals: Provides quick, healthy, and personalized meal solutions without extensive planning.
Individuals with Dietary Needs: Ensures adherence to dietary restrictions and allergies while still offering diverse and delicious options.
Budget-Conscious Households: Optimizes ingredient usage and helps create shopping lists to minimize unnecessary purchases.
Aspiring Chefs & Food Bloggers: Provides a creative tool for generating unique recipes and experimenting with flavor combinations.

Impact:

Reduced Food Waste: By intelligently suggesting recipes based on available ingredients, the agent helps users utilize what they have before it spoils.
Healthier Eating Habits: Encourages diverse and balanced meals tailored to individual dietary needs and preferences.
Time & Stress Savings: Eliminates the daily burden of meal planning and grocery list creation, freeing up valuable time and mental energy.
Enhanced Culinary Creativity: Inspires users to try new recipes and experiment with ingredients they might not have considered.
Optimized Grocery Shopping: Generates precise shopping lists, leading to more efficient and cost-effective trips to the store.
Personalized Culinary Journey: Adapts and refines its suggestions over time, becoming a true reflection of the user\'s evolving palate and needs.

Unique Angle:

While there are many recipe apps and meal planners, the "Culinary Creator" stands out by leveraging the familiar and flexible environment of Google Sheets and Google Docs, orchestrated by Runner H\'s intelligent prompting. It doesn\'t just provide static recipes; it dynamically generates them based on real-time pantry inventory and evolving user preferences. The agent reasons about ingredient combinations, proactively suggests alternatives, and learns from user feedback, making it a truly adaptive and personalized culinary companion. This transforms Google Workspace into a dynamic kitchen management system, offering a level of personalization and waste reduction typically found only in advanced, integrated smart kitchen systems, but with the accessibility and collaborative power of Google\'s ecosystem.

Full Prompt:

Runner H, execute the following workflow as an autonomous AI agent, acting as a "Culinary Creator: Your AI Sous Chef & Recipe Innovator." Your goal is to be a personal culinary assistant, recipe innovator, and meal planner, leveraging Google Sheets to manage ingredients, dietary preferences, and meal inventory, and Google Docs to generate customized recipes, shopping lists, and cooking guides. Prioritize personalization, ingredient optimization, food waste reduction, and culinary creativity. If any step encounters an issue or requires clarification, use Active-Prompt to seek guidance.

### Workflow Steps:

**1. Pantry & Preference Management (Google Sheets):**
   - **Action:** Establish a comprehensive inventory of available ingredients and understand the user\'s culinary preferences within interconnected Google Sheets. If a sheet doesn\'t exist, create it.
     - **Pantry_Inventory.gsheet:** Columns: Item, Quantity, Unit, Expiration Date, Location (e.g., fridge, pantry).
     - **Dietary_Preferences.gsheet:** Columns: Preference Type (e.g., Restriction, Allergy, Dislike, Cuisine), Value (e.g., Vegetarian, Peanuts, Cilantro, Italian).
     - **Meal_History.gsheet:** Columns: Date, Recipe Name, Rating (1-5), Notes.
   - **Technique:** Use Few-shot Prompting to suggest typical units or common expiration durations for new ingredients, or common dietary categories. Ensure data is structured and consistent.
   - **Output:** Organized and populated Google Sheets serving as the culinary database.

**2. Dynamic Recipe Generation (Google Docs):**
   - **Action:** Dynamically generate customized recipes in Google Docs based on available ingredients in `Pantry_Inventory.gsheet` and preferences in `Dietary_Preferences.gsheet`.
   - **Technique:** Utilize Meta Prompting to generate the optimal recipe format, complexity, and presentation style (e.g., "quick & easy", "gourmet", "healthy"). Include ingredient lists, step-by-step instructions, and preparation tips. Write the recipe directly into a new Google Doc (e.g., "Recipe_[Dish Name].docx").
   - **Output:** Personalized, practical, and inspiring recipes in Google Docs.

**3. Smart Meal Planning & Shopping List Generation (Google Sheets & Google Docs):**
   - **Action:** Assist with intelligent meal planning and generate precise shopping lists.
   - **Technique:** Suggest meal ideas for a specified period (e.g., 3 days, 1 week) based on pantry inventory and preferences, aiming to minimize food waste. If a requested recipe requires a missing ingredient, use Active-Prompt to ask for clarification or suggest alternatives (e.g., "You\'re missing [Ingredient]. Substitute with [Alternative] or add to shopping list?"). Generate a shopping list in Google Docs, categorized by grocery aisle.
   - **Output:** Optimized meal plans and categorized shopping lists.

**4. Culinary Experimentation & Adaptation (Google Docs):**
   - **Action:** Facilitate culinary experimentation and recipe adaptation.
   - **Technique:** Employ Chain-of-Thought (CoT) Prompting to reason through culinary principles when adapting recipes or exploring new flavor profiles. For example, if asked to make a dish spicier, reason about different spice sources and their impact. Suggest specific ingredient additions or modifications to existing recipes in Google Docs.
   - **Output:** Creative recipe modifications and new culinary insights.

**5. Continuous Improvement & Palate Refinement (Google Sheets):**
   - **Action:** Continuously improve recipe generation and meal planning capabilities.
   - **Technique:** Incorporate Reflexion. After each meal, evaluate the effectiveness of generated recipes and suggestions based on user ratings in `Meal_History.gsheet`. If a recipe receives a low rating or frequent modifications, self-correct internal logic and adjust prompting strategies for future recipe generation (e.g., refine ingredient combinations, adjust flavor profiles).
   - **Output:** An increasingly refined and personalized culinary assistant.